This paper describes a syntactic approach to deducing the logical structure of printed documents from their physical layout. Page layout is described by a 2-dimensional grammar, similar to a context -free string grammar, and a chart parser is used to parse segmented page images according to the grammar. This process is part of a system which reads scanned document images and produces computer readable text in a logical mark-up format such as SGML. The system is briefly outlined, the grammar formalism and the parsing algorithm are described in detail, and some experimental results are reported. and semantic information is used to link the document into a hypertext network. Mark-up helps the indexing process by highlighting information rich arm such as the abstract, and also provides structural links for hypertext browsing.

The sign languages used by deaf communities around the world represent a linguistic challenge that natural language researchers have only recently begun to take up. Zardoz is a system which tackles the cross-modal machine-translation problem, translating speech and text into animated sign language. Native sign !languages, such as ISL (Ireland), BSL (Britain) and ASL (U.S.A.) have evolved in deaf communities as natural methods of gestural communication. These languages differ from English, not only in modality, but in grammatical structure, exploiting the dimensions of space as well as time. This paper presents an architectural overview of Zardoz, and describes the methods employed to analyse the verbal input and generate the corresponding signed output.

No abstract

No abstract